Abstract: The geographical features,i.e.latitude,altitude,distance to large water area and mountain structure,etc.,are commonly accepted as the global factors impacting the spatial distribution of groud surface temperature.However in the mountainous area,local topographical features,such as slope gradient,aspect and terrain shading play a significant role to the redistribution of thermal energy.The traditional method for calculating the spatial distribution of surface temperature in rugged areas usually employs a model of statistic regression,namely a regression of temperature against latitude,longitude and altitude,etc.,which is suitable for revealing the regional surface temperature,but not for refined climatic source exploration and application.Taking Yaoxian County in Shaanxi Province as a test area,DEM dataset of 1:50000 map as basic information sources,the authors conducted a terrain-revised model for calculating temperature at local slope level.The key process is a cell by cell revising of traditional model by calculating the astronomical radiation based on DEM.A comparison between simulated temperature and radiation one from TM6 thermal infrared image shows that the new method gets more reasonable result which could reveal not only accurate local temperature,but also their finer distribution.It will be great helpful to improve the temperature simulation both qualitatively and effectively.